Leveraging Smart Wearable Internet of Things Systems for Remote Healthcare Monitoring Using Dimensionality Reduction with Deep Representation Learning Model

Technical and accessibility concerns in hospitals frequently prevent individuals from receiving the best physical and mental health care, which is vital for independent living. Current advancements in the Internet of Things (IoT)-based wearable devices provide capable applications for remote health monitoring and day-to-day activity detection, which has attained substantial interest in healthcare. Furthermore, the integration of wearables in the IoT system allows healthcare systems to use interconnected technologies, thus raising treatment protocols, enhancing diagnostic precision, and improving healthcare delivery and patient outcomes. Body wearable devices have gained attention as robust devices for healthcare applications, resulting in numerous commercially accessible devices for several motives, such as personalized healthcare, activity alerts, and fitness. Deep learning (DL) holds a prominent position in transforming remote healthcare by improving diagnostic precision, allowing real-time monitoring, and enabling tailored treatment plans remotely. Moreover, DL’s prediction abilities are revolutionizing patient monitoring and preventive care in remote healthcare environments. This paper presents an Intelligent Remote Healthcare Monitoring Framework Using Feature Selection and Deep Autoencoder (IRHMFS-DAE) model using IoT-integrated wearable devices. The objective of this framework is to provide a promising approach for proactive disease detection and personalized healthcare management. In the data preprocessing stage, the proposed IRHMFS-DAE method applies data cleaning and normalization to enhance the quality of the collected sensor signals and ensure consistency. Next, several dimensionality reduction techniques are utilised for identifying the most relevant health-related features and enhancing interpretability for medical decision-making. For healthcare data classification, a stacked denoising autoencoder (SDAE) is deployed to effectively learn complex patterns in patient data and to enhance prediction accuracy. To determine the heightened performance of the IRHMFS-DAE model, numerous simulations were performed, and the results were inspected under several measures. The comparison investigation stated the improvement of the IRHMFS-DAE method under diverse measures.